Review of Automated Vulnerability Analysis of Smart Contracts on Ethereum

Abstract

Programs on public blockchains often handle valuable assets, making them attractive targets for attack. At the same time, it is challenging to design correct blockchain applications. Checking code for potential vulnerabilities is a viable option to increase trust. Therefore, numerous methods and tools have been proposed with the intention to support developers and analysts in detecting code vulnerabilities. Moreover, publications keep emerging with different focus, scope, and quality, making it difficult to keep up with the field and to identify relevant trends. Thus, regular reviews are essential to keep pace with the varied developments in a structured manner. Regarding blockchain programs, Ethereum is the platform most widely used and best documented. Moreover, applications based on Ethereum are entrusted with billions of USD. Like on similar blockchains, they are subject to numerous attacks and losses due to vulnerabilities that exist at all levels of the ecosystem. Countermeasures are in great demand. In this work, we perform a systematic literature review (SLR) to assess the state of the art regarding automated vulnerability analysis of smart contracts on Ethereum with a focus on classifications of vulnerabilities, detection methods, security analysis tools, and benchmarks for the assessment of tools. Our initial search of the major on-line libraries yields more than 1,300 publications. For the review, we apply a clear strategy and protocol to assure consequent, comprehensive, and reproducible documentation and results. After collecting the initial results, cleaning up references, removing duplicates and applying the inclusion and exclusion criteria, we retain 303 publications that include 214 primary studies, 70 surveys and 19 SLRs. For quality appraisal, we assess their intrinsic quality (derived from the reputation of the publication venue) as well as their contextual quality (determined by rating predefined criteria). For about 200 publications with at least a medium score, we extract the vulnerabilities, methods, and tools addressed, among other data. In a second step, we synthesize and structure the data into a classification of both the smart contract weaknesses and the analysis methods. Furthermore, we give an overview of tools and benchmarks used to evaluate tools. Finally, we provide a detailed discussion.